Intuitive fuzzy-fused subjective and objective dynamic balance seaborne-to-sea decision scheme evaluation method

Abstract

The invention belongs to the field of data analysis and evaluation of decision-making schemes, the field of machine learning and cognitive calculation and the field of intelligent decision-making, and relates to a subjective and objective dynamic balance seaborne-to-sea decision-making scheme evaluation method fusing intuitionistic fuzzy. The method comprises the following steps: firstly, simulating a subjective dimension by utilizing an intuitionistic fuzzy set IFSs to obtain an auxiliary quantitative index; secondly, accurately calculating objective weights of the quantitative indexes and the auxiliary indexes thereof by adopting an entropy weight method; the subjective weight of the qualitative index is calculated by using the Bayesian network, so that the interpretability and accuracy of subjective judgment are ensured; and finally, organically combining subjective and objective weights through a dynamic combination mode, and calculating a comprehensive score of the sea-to-sea decision scheme. The method not only overcomes the defect that subjective factors are too heavy in traditional evaluation, but also fully considers objective actual conditions, and can realize scientific and comprehensive evaluation of the sea-to-sea decision scheme.

Description

Technical Field

[0001] The present invention belongs to the fields of data analysis and evaluation of decision-making schemes, machine learning and cognitive computing, and intelligent decision-making, and relates to a method for evaluating maritime decision-making schemes that integrates subjective and objective dynamic balance with intuitive fuzzy. Background Art

[0002] The maritime sea-to-sea decision-making plan is a systematic sea-to-sea mission planning system, whose core goal is to strike surface ship targets in the sea direction. The planning process begins with clarifying the decision-making goal, and then comprehensively considers factors such as cost, time, resources and effect, and finally forms a variety of feasible sea-to-sea decision-making plans. The evaluation of maritime sea-to-sea decision-making plans is a comprehensive decision-making support process. By collecting key data, building mathematical models, conducting computer simulations, and applying multi-criteria decision analysis technology, a comprehensive evaluation and selection of various maritime sea-to-sea decision-making plans can be achieved. In the evaluation process of maritime sea-to-sea decision-making plans, key factors such as cost, time, resources and effect must be comprehensively considered to ensure that commanders make scientific and reasonable decisions in the face of uncertainty and resource constraints.

[0003] At present, the mainstream evaluation methods for maritime decision-making schemes mainly rely on new technologies in the fields of computers and mathematics, including statistical probability models, computer simulation, multiple criteria decision making (MCDM), artificial intelligence and machine learning algorithms, etc. These methods play a key role in the evaluation of maritime decision-making schemes and provide strong support for decision makers. However, there is still a long way to go before the truly systematic and scientific evaluation of maritime decision-making schemes. Insufficient mining of objective information, excessive intervention of subjectivity, and lack of dynamic adaptability may affect the accuracy of the evaluation. Therefore, the evaluation method needs to be further developed to enhance its ability to process complex data, reduce subjective bias, and provide clear decision-making basis under information uncertainty and dynamic changes. Specifically, the current evaluation methods and frameworks for maritime decision-making schemes need to solve the following problems:

[0004] 1. The objective weight calculation of quantitative indicators is unreasonable: Due to the limitation of data quality, the limitation of analysis methods and the inaccuracy of information, the mining and analysis of objective information is not in-depth enough, and the objective weight calculation of quantitative indicators is unreasonable, which leads to the deviation of evaluation results.

[0005] 2. The subjective weight of qualitative indicators is overly dependent on experience: Qualitative indicators (such as reliability, compatibility, hit rate, damage effect, etc.) are difficult to quantify. Decision makers may be influenced by personal preferences, intuition or past experience when assigning weights, thereby introducing subjective biases and affecting the objectivity and universality of the decision.

[0006] 3. Imbalance in subjective and objective weighting: In the weight allocation of maritime decision-making plans, decision makers may tend to rely on subjective judgment and ignore objective data analysis, or rely too much on data and ignore intuition and experience, resulting in an imbalance in the allocation of subjective and objective weights. This imbalance affects the comprehensiveness and accuracy of decision-making, thus causing the decision results to deviate from the optimal solution.

[0007] Compared with the existing marine decision-making scheme evaluation, the innovation and advantages of the present invention are as follows:

[0008] 1. Use intuitionistic fuzzy sets to generate auxiliary quantitative indicators, enhance the subjectivity of quantitative indicators by the ambiguity of the affiliation of the quantitative element x, and fully explore the qualitative characteristics of quantitative indicators in the evaluation of maritime decision-making plans.

[0009] 2. The Entropy Weight Method (EWM) is used to integrate and calculate the quantitative indicators and auxiliary quantitative indicators, further explore the objective and subjective information in the quantitative indicators, and improve the rationality of the objective weight calculation.

[0010] 3. Apply Bayesian networks to calculate the subjective weights of qualitative indicators. By constructing models and probability tables, an objective and formal description of qualitative indicators in the evaluation of maritime decision-making plans can be achieved, the problem of characterizing the ambiguity of qualitative indicators can be solved, and the reliability of subjective weight calculation can be improved.

[0011] 4. Construct a calculation method for the combined weight index that combines subjective and objective factors, and comprehensively consider the impact of subjective and objective factors on the decision-making plan at sea. Based on the established subjective and objective weighting principles and clear judgment criteria, the assignment of subjective and objective weights is adaptively and dynamically adjusted to ensure that the final score can more reasonably reflect the impact of subjective and objective factors. Summary of the invention

[0012] The technical problems to be solved by the present invention are:

[0013] (1) Problems in calculating quantitative weights due to insufficient objective information: There is a need to improve the problem of unreasonable weight allocation of quantitative indicators due to insufficient utilization of objective information.

[0014] (2) The problem of qualitative weight calculation caused by experience dependence: This paper solves the problem of insufficient interpretability and reliability of qualitative indicator weight calculation due to excessive reliance on personal experience.

[0015] (3) Lack of dynamic adjustment of subjective and objective weights: A mechanism needs to be developed to reasonably consider and dynamically adjust the proportion of subjective and objective weights to prevent decision-making results from deviating from objective facts.

[0016] The technical solution of the present invention is as follows:

[0017] The evaluation method of maritime decision-making scheme integrating subjective and objective balance with intuitionistic fuzzy is as follows:

[0018] Step S1: construct an evaluation index system for the decision-making scheme on the sea, clearly distinguish quantitative indicators from qualitative indicators, and establish corresponding theoretical models for these indicators;

[0019] Step S2: construct an evaluation matrix, collect evaluation index-related data, and perform range standardization on quantitative index data to eliminate the dimensional effects between different indicators;

[0020] Step S3: Use intuitionistic fuzzy sets (IFSs) to pseudo-subjectively process quantitative indicators, and calculate objective weights using the entropy weight method;

[0021] Step S4: Apply the Bayesian network to model the qualitative indicators and calculate their subjective weights;

[0022] Step S5: Using a dynamic combination method to integrate the subjective and objective weights to achieve a dynamic balance of weights;

[0023] Step S6: score each sea decision-making scheme at sea according to the calculated combined weight, and rank them according to their merits, and finally obtain the evaluation results of each sea decision-making scheme at sea.

[0024] Step S1, construct an evaluation index system for maritime decision-making schemes and model each evaluation index: Let T = {t 1 ,t 2 …,t a} are a experts who evaluate the decision-making plan for the sea. Let S = {s 1 ,s 2 ,…,s b} is a set of b maritime decision-making plans. According to the key elements of the maritime decision-making plans, an evaluation grade classification standard set V = {0, 1, 2, 3, 4, 5} is established, where the larger the value, the greater the weight. The evaluation object is the evaluation index in the pre-compiled c maritime decision-making plans. Let the evaluation index set U = {u 1 ,u 2 ,…,u c}.

[0025] Step S2, construct an evaluation matrix for the sea-to-sea decision-making scheme: Based on the evaluation index model established in step S1, construct an evaluation matrix M for each sea-to-sea decision-making scheme. The matrix is ​​a two-dimensional array, M has b rows, each row represents a different evaluation scheme, corresponding to the sea-to-sea decision-making scheme set S, M has c columns, each column corresponds to a different evaluation index, corresponding to the evaluation index set U. For the quantitative index set, collect data related to each quantitative evaluation index. These data may come from actual data, expert evaluation, experimental training, simulation deduction or intelligence analysis.

[0026] Taking the launch time and launch position as an example, first, the test training data of these indicators are sorted out. For the data of quantitative indicators, the numerical data are classified and filled into the evaluation matrix according to the evaluation indicator set U; for the data of qualitative indicators, according to the evaluation grade grading standard set V, the numerical scores of a expert evaluation opinions in T are obtained, and these numerical data are filled into the evaluation matrix. Finally, evaluation matrices are generated for multiple maritime decision-making plans to obtain a evaluation matrices.

[0027]

[0028] Among them, i∈[1,a], score∈V.

[0029] Step S3, pseudo-subjective quantitative indicators, and calculate the objective weights through the entropy weight method: According to the evaluation matrix constructed in step S2, the quantitative indicators are pseudo-subjectively calculated through intuitive fuzzy sets IFSs to obtain auxiliary quantitative indicators (pseudo-qualitative indicators), and the entropy weight method (Entropy Weight Method, EWM) is used to fuse the quantitative indicators and auxiliary quantitative indicators to achieve reasonable calculation of objective weights. The addition of auxiliary quantitative indicators makes the quantitative indicators subjective, thereby more effectively exerting the information value of the quantitative indicators.

[0030] Specifically, the data set X=U is defined as the set of elements that may appear in the indicators considered in all maritime decision-making plans. This data set provides a complete and non-missing set of elements for the intuitionistic fuzzy set.

[0031] Furthermore, we define an intuitionistic fuzzy set A. Based on the data set X, A can be expressed as: A = {(x, μ A (x),γ A (x))|x∈X}. Among them, μ A (x) is the membership function, which indicates the degree to which element x belongs to set A; γ A (x) is the non-membership function, which indicates the degree to which element x does not belong to set A.

[0032] Among them, the membership degree μ A (U) and non-membership γA (U) It needs to be determined through expert knowledge or experience, or it can be determined based on preset rules of expert knowledge.

[0033] Furthermore, through π A (x) = 1-μ A (x)-γ A (x) Calculate the uncertainty measure. The higher the value obtained, the higher the uncertainty about whether element x belongs to set A. The uncertainty measure quantifies the ambiguity of the affiliation of element x and expresses the uncertainty of the indicator in numerical form. A high uncertainty measure means that there is a greater hesitation or uncertainty in the classification of element x, that is, the pseudo-subjectivity of element x (i.e., quantitative indicator) is higher. Pseudo-subjectivity means that during the evaluation process, a certain degree of subjectivity is given to those quantitative indicators that are objective in nature through specific methods in order to better reflect the experience, intuition and preferences of decision makers. The addition of pseudo-qualitative indicators makes quantitative indicators subjective, thereby more effectively exerting the information value of quantitative indicators and achieving the purpose of fully mining the information of quantitative indicators.

[0034] Furthermore, a mapping from quantitative indicators to intuitionistic fuzzy set A is defined. Based on membership and non-membership, the quantitative indicators are converted into elements in intuitionistic fuzzy set A, and the objective weights of the quantitative indicators are calculated through the entropy weight method.

[0035] Specifically, the entropy weight method is used to calculate the objective weight:

[0036] Since the dimensions and numerical ranges of different indicators may be different, the original data needs to be standardized first. Standardize the quantitative indicators and auxiliary quantitative indicators. ij is the original data of the jth indicator under the i-th scheme, x i ′ j is the standardized data, max(x i ) and min(x i ) are the maximum and minimum values ​​of all schemes of the ith indicator. The range standardization method can simply, intuitively and quickly convert data in different dimensions and numerical ranges into a unified [0,1] interval, which is convenient for subsequent comparison and processing.

[0037] Furthermore, the information entropy that measures the discreteness of quantitative indicators and auxiliary quantitative indicator data is calculated. Among them, e j is the information entropy of the jth indicator, b is the number of solutions, p ij is the standardized data x i ′ jThe normalized value of is calculated as The degree of dispersion is used to measure the diversity and uncertainty of the distribution between indicators. A low degree of dispersion means that the indicators are more dispersed within the possible range of values, while a high degree of dispersion means that the data values ​​tend to be concentrated.

[0038] Taking the launch time and launch position indicators as an example, the time and position data with concentrated discrete degree areas are identified as repeated objective indicators and are assigned higher discrete values.

[0039] Furthermore, the coefficient of difference g used to measure the ability of the indicator to distinguish is calculated j =1-e j Among them, g j is the coefficient of variation of the jth indicator. The coefficient of variation is used to quantify the relative dispersion of indicator data and reflects the comparison between the centralized trend and the discrete trend of the data set.

[0040] Taking the launch time and launch position indicators as an example, if the launch time indicator has a higher coefficient of variation, it means that the indicator has a greater variation than the launch position indicator and has a stronger ability to distinguish different situations.

[0041] Finally, the weight of each indicator is calculated based on the coefficient of variation The larger the weight, the stronger the distinguishing ability of the indicator is, and the more important it is to the objective matrix. j is the weight of the jth indicator, and c is the total number of indicators.

[0042] Step S4, using the Bayesian network to calculate the subjective weight of the qualitative indicators:

[0043] First, construct a Bayesian network graph structure, namely a directed acyclic graph (DAG). Each node represents the score of a certain indicator in different schemes, and the edge represents the correlation between qualitative indicators. The network structure G can be expressed as: G = (V, E). Among them, V is the node set and E is the edge set.

[0044] Specifically, in order to analyze the inherent fuzziness and uncertainty in qualitative indicators, a conditional probability table (CPT) is defined for each variable in the network to describe the probability distribution of the corresponding variable when the value of the relationship variable is given. f (X), CPT can be expressed as: P(X|P f (X)). Where P(X|P f (X)) is the relation variable P f (X) The probability distribution of variable X under given conditions.

[0045] Furthermore, the evaluation expert knowledge and relationship indicators are used to learn the parameters in the CPT. It should be noted that if the variable X has n possible values, its CPT will contain n probability values, which need to satisfy the normalization condition of probability: ∑ x P(X=x|P f (X)) = 1. Here, x traverses all possible values ​​of X.

[0046] Specifically, the expectation-maximization (EM) algorithm is used to learn the network parameters.

[0047] First, initialize a θ parameter.

[0048] Next, with the current parameter θ t and observed variable X to infer latent variable x to obtain distribution P(x|X,θ t ), and calculate the log-likelihood estimate LogL(θ t |X,x)Expectation about x:

[0049] Find the parameter that maximizes the expected likelihood, that is: θ t+1 = arg maxQ(θ|θ t ).

[0050] Finally, the above two steps are repeated until the set convergence threshold or number of iterations is reached to obtain the local or global optimal parameters.

[0051] Furthermore, the constructed Bayesian network is used for reasoning. Using the learned conditional probability table P(X|P f (X)), through the formula The subjective weight is calculated. Where X is the relationship variable of Y, and P(E) is the probability of given condition E.

[0052] Step S5, using a dynamic combination method to calculate the subjective and objective dynamic combination weights:

[0053] Using the formula Calculate the dynamic combination weight. Where W j is the comprehensive weight of the jth indicator, W j ′ To calculate the subjective weight of the jth indicator through the Ye Si network, W j″ is the objective weight of the jth indicator calculated by the intuitionistic fuzzy entropy weight method, and c is the number of evaluation indicators. Compared with the traditional simple combination method, the formula uses a normalized method to dynamically balance the subjective and objective weights. Relatively high comprehensive weights are assigned to indicators with a large proportion, and relatively low comprehensive weights are assigned to indicators with a small proportion. This helps to avoid the problem of scale inconsistency in weight calculation, making weight comparison and interpretation more intuitive and fair, thereby reasonably representing the impact of each indicator on the overall decision in subjective and objective multi-criteria decision analysis problems such as the evaluation of maritime decision-making plans.

[0054] Step S6, obtaining a final score according to the combined weight, ranking the multiple sea-going decision-making schemes by their merits, and obtaining the evaluation results of the multiple sea-going decision-making schemes:

[0055] Using weighted composite operation The comprehensive score of the decision-making plan on the sea is obtained. i is the comprehensive score of the i-th maritime decision-making plan, W j is the comprehensive weight of the jth indicator, x j is the score of the jth indicator, and c is the number of indicators.

[0056] Execute the sorting algorithm according to the comprehensive score R i , sort all the sea-related decision-making options, and generate a list R={…,s i ,...}.

[0057] Beneficial effects of the present invention:

[0058] The present invention aims at the unreasonable calculation of objective weights of quantitative indicators in the field of maritime decision-making scheme evaluation, the excessive reliance on experience of subjective weights of qualitative indicators, and the imbalance of subjective and objective weights in the field of maritime decision-making scheme evaluation. A subjective and objective dynamic balance maritime decision-making scheme evaluation method integrating intuitive fuzzy sets is proposed. In terms of specific implementation, the method first uses intuitive fuzzy sets IFSs to simulate the subjective dimension and obtain auxiliary quantitative indicators; then the entropy weight method is used to accurately calculate the objective weights of these quantitative indicators and their auxiliary indicators; then the Bayesian network is used to calculate the subjective weights of qualitative indicators to ensure the interpretability and accuracy of subjective judgment; finally, through a dynamic combination method, the subjective and objective weights are organically combined to calculate the comprehensive score of the maritime decision-making scheme. The proposed method not only overcomes the deficiency of excessive subjective factors in traditional evaluation, but also fully considers the objective actual situation, and can realize the scientific and comprehensive comprehensive evaluation of maritime decision-making schemes, which provides strong support for commanders to analyze and predict the feasibility and effectiveness of decision-making schemes, and better manage and supervise decision-making schemes. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 is a flow chart of the present invention;

[0060] Figure 2 This is a flow chart of objective weight calculation of the present invention;

[0061] Figure 3 Flowchart for subjective weight calculation;

[0062] Figure 4 A diagram of a directed acyclic graph. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0064] Assuming that there is friction between the Blue and Red sides at sea, the Red side decides to take necessary measures to attack in order to safeguard its maritime rights and interests. The maritime command post plans to select the best maritime decision-making plan from five alternative maritime decision-making plans. An evaluation index system for maritime decision-making plans is established in accordance with the principles and general process of establishing an evaluation index system for maritime decision-making plans. By integrating the opinions of five experts, six evaluation indicators for maritime decision-making plans are summarized and designed.

[0065] First, the various components of the evaluation method are modeled. From the above description, we can get: the evaluation expert set T = {t 1 ,t 2 ,t 3 ,t 4 ,t 5}, the decision-making solution set S = {s 1 ,s 2 ,s 3 ,s 4 ,s 5}, evaluation index set U = {u 1 ,u 2 ,u 3 ,u 4 ,u 5 ,u 6 In addition, the evaluation grade grading standard set V = {0, 1, 2, 3, 4, 5}.

[0066] Based on the various components of the evaluation method obtained above, an evaluation matrix set of the sea decision-making plan is constructed: m = {M 1 ,M 2 ,M 3,M 4 ,M 5 The evaluation index set U = {hitting capability, launching time, launching interval, coordinated attack, damage capability, launching position}, launching time and launching position are quantitative indicators, and hitting capability, launching interval, coordinated attack, and damage capability are qualitative indicators.

[0067] Table 1 Evaluation matrix set of sea-related decision-making options

[0068]

[0069] M 1 Take the auxiliary objective matrix weight as an example:

[0070] Define the data set X = U = {u 1 ,u 2 ,u 3 ,u 4 ,u 5 ,u 6}.

[0071] The membership and non-membership are determined through the experience of experts, and the corresponding uncertainty measure is determined through the formula to further calculate the intuitive fuzzy set: μ A (u 1 )=0.8,γ A (u 1 )=0.1,π A (u 1 )=0.1,μ A (u 2 )=0.7,γ A (u 2 )=0.2,π A (u 2 )=0.1,μ A (u 3 )=0.6,γ A (u 3 )=0.3,π A (u 3 )=0.1,μ A (u 4 )=0.8,γ A (u 4 )=0.1,π A (u 4 )=0.1,μ A (u 5 )=0.7,γ A (u 5 )=0.2,π A (u 5 )=0.1,μ A (u 6)=0.9,γ A (u 6 )=0.05,π A (u 6 )=0.05.

[0072] Table 2 Evaluation matrix M for sea-related decision-making options at sea 1 Quantitative indicator intuitionistic fuzzy sets

[0073]

[0074] Define the mapping y = x*μ A (x)+γ A (x) Obtain the auxiliary objective matrix M 1 ′:

[0075] Table 3. Evaluation matrix of sea decision-making scheme at sea. Auxiliary objective matrix M 1 '

[0076]

[0077]

[0078] The objective weight is calculated using the EAM entropy weight method:

[0079] M 1 ′ as an example, the objective matrix and the auxiliary objective matrix are standardized using the range standardization formula:

[0080] Table 4 Evaluation matrix of sea decision-making scheme at sea Auxiliary objective matrix Standardization M 1 ″

[0081]

[0082] According to the information entropy formula, the information entropy and difference coefficient are calculated: e(u 2 )≈0.80, g(u 2 )=1-e(u 2 )=0.2,e(u 6 )≈0.86, g(u 6 )=1-e(u 6 )=0.14.

[0083] Calculate the auxiliary objective weight matrix according to the objective weight formula:

[0084] Table 5. Auxiliary objective weight matrix M of the evaluation matrix of the decision-making scheme for sea 1 ″′

[0085]

[0086] In the same way, we obtain the objective matrix M1″″:

[0087] Table 6 Objective weight matrix M of the evaluation matrix of the decision-making scheme for sea 1 ″″

[0088]

[0089] Get the final objective weight M1″″′:

[0090] Table 7 Objective weight matrix M of the evaluation matrix of the decision-making scheme for sea 1 ″″′

[0091]

[0092] M 1 As an example, use the Bayesian network to calculate the subjective weight matrix:

[0093] Table 8 Evaluation matrix M for sea decision-making schemes at sea 1 Qualitative indicators

[0094]

[0095]

[0096] Construct a directed acyclic graph DAG for the qualitative indicator matrix, such as Figure 4 shown.

[0097] Define conditional probability based on expert knowledge:

[0098] Assume that each indicator has only two states: high (H) and low (L). 1 )=0.5(high)0.5(low, P(u 3 ∣u 1 )=0.7(high|u 1 High)0.3(High|u 1 Low), P(u 4 ∣u 1 )=0.8(high|u 1 High)0.2(High|u 1 Low), P(u 5 ∣u 3 ,u 4 )=0.9(high|u 3 High and u 4 High)0.1(High|Other cases).

[0099] with u 1 Take this as an example to calculate the degree of influence of other indicators under given conditions:

[0100] Calculate P(u 3 ∣u 1 ), that is, considering u1 Under the given conditions u 3 The probability of being affected is high.

[0101] u 1 and u 3 The relationship is unidirectional, that is, u 3 Directly affected by 1 Influence, and u 1 not subject to u 3 Therefore, you can directly use u 1 When it happens 3 is a high conditional probability, that is, P(u 3 = high |u 1 = high) = 0.7.

[0102] Calculate P(u 4 ∣u 1 ), that is, considering u 1 Under the given conditions u 4 The probability of being affected is high.

[0103] According to the definition of conditional probability, we can directly use u 1 When it happens 4 is a high conditional probability, that is, P(u 4 = high |u 1 = high) = 0.8.

[0104] Calculate P(u 5 ∣u 1 ), that is, considering u 1 Under the given conditions u 5 The probability of being affected is high.

[0105] Using the total probability formula, we can write:

[0106] P(u 5 =H) = ∑P(u 5 =H|u 3 ,u 4 )·P(u 3 )·P(u 4 )

[0107] Because u 3 and u 4 All depend on u 1 , the above formula can be expanded as:

[0108] P(u 5 =H) = P(u 5 =H|u 3 =H,u 4 =H)·P(u 3 =H|u 1)·P(u 4 =H|u 1 )·P(u 1 )+P(u 5 =H|u 3 =L,u 4 =H)·P(u 3 =L|u 1 )·P(u 4 =H|u 1 )·(1-P(u 1 ))+P(u 5 =H|u 3 =H,u 4 =L)·P(u 3 =H|u 1 )·P(u 4 =L|u 1 )·P(u 1 )+P(u 5 =H|u 3 =L,u 4 =L)·P(u 3 =L|u 1 )·P(u 4 =L|u 1 )·(1-P(u 1 ))

[0109] Substituting the probability value, we get: P(u 5 =H) = 0.3084

[0110] So we get u 1 The degree of influence of other indicators under given conditions:

[0111] Table 9 Qualitative indicators u 1 Impact on other indicators

[0112]

[0113] According to the above steps, the impact of other indicators is calculated as follows:

[0114] Table 8 Evaluation matrix M for sea decision-making schemes at sea 1 Qualitative indicators affected

[0115]

[0116]

[0117] After normalization, the final subjective weight is obtained:

[0118] Table 10 Subjective weight matrix of the evaluation matrix of the decision-making scheme for sea operations

[0119]

[0120] The combination weight is calculated according to the subjective and objective dynamic combination weight formula:

[0121] Table 10 Decision-making options for sea operations at sea M 1 Evaluation matrix combination weight matrix

[0122]

[0123] The final score of M1 is obtained by using the weighted synthesis calculation formula:

[0124] Table 11 Decision-making options for sea operations at sea M 1 Final Rating

[0125]

[0126] The ranking R is obtained as {2, 1, 3, 4, 5}.

[0127] Repeat the above steps to calculate M 2 To M 5 The final score is obtained to obtain the final ranking.

Claims

1. A subjective and objective balance maritime decision-making scheme evaluation method integrating intuitive fuzzy, characterized by: Here are the steps: Step S1: construct an evaluation index system for the decision-making scheme on the sea, clearly distinguish quantitative indicators from qualitative indicators, and establish corresponding theoretical models for these indicators; Step S2: construct an evaluation matrix, collect evaluation index-related data, and perform range standardization on quantitative index data to eliminate the dimensional effects between different indicators; Step S3: Use intuitionistic fuzzy sets (IFSs) to pseudo-subjectively process quantitative indicators, and calculate objective weights using the entropy weight method; Step S4: Apply the Bayesian network to model the qualitative indicators and calculate their subjective weights; Step S5: Using a dynamic combination method to integrate the subjective and objective weights to achieve a dynamic balance of weights; Step S6: score each sea decision-making scheme at sea according to the calculated combined weight, and rank them according to their merits, and finally obtain the evaluation results of each sea decision-making scheme at sea.

2. The subjective and objective balance maritime decision-making scheme evaluation method integrating intuitionistic fuzzy as claimed in claim 1 is characterized in that: The specific operations of step S1 are as follows: Construct an evaluation index system for maritime decision-making schemes and model each evaluation index: Let T = {t1, t2…, t a } are a experts who evaluate the decision-making plan at sea; let S = {s1,s2,…,s b } is a set of b maritime decision-making schemes; according to the key elements of the maritime decision-making schemes, an evaluation grade classification standard set V = {0, 1, 2, 3, 4, 5} is established, where the larger the value, the greater the weight; the evaluation object is the evaluation indicators in the pre-compiled c maritime decision-making schemes, and the evaluation indicator set U = {u1, u2, …, u c }.

3. The subjective and objective balance maritime decision-making scheme evaluation method integrating intuitionistic fuzzy as claimed in claim 1 is characterized in that: The specific operation of step S2 is as follows: Constructing an evaluation matrix for the maritime sea decision-making scheme: Based on the evaluation index model established in step S1, construct an evaluation matrix M for each maritime sea decision-making scheme; the matrix is ​​a two-dimensional array, M has b rows, each row represents a different evaluation scheme, corresponding to the maritime sea decision-making scheme set S; M has c columns, each column corresponds to a different evaluation index, corresponding to the evaluation index set U; for the quantitative index set, collect data related to each quantitative evaluation index; these data may come from actual data, expert evaluation, experimental training, simulation deduction or intelligence analysis; Taking the launch time and launch position as an example, first, the test training data of these indicators are sorted out. For the data of quantitative indicators, the numerical data are classified and filled into the evaluation matrix according to the evaluation indicator set U; for the data of qualitative indicators, the numerical scores of a expert evaluation opinions in T are obtained according to the evaluation grade classification standard set V, and these numerical data are filled into the evaluation matrix. Finally, the evaluation matrix is ​​generated for multiple maritime decision-making schemes to obtain a evaluation matrix; Among them, i∈[1,a], score∈V.

4. The subjective and objective balance maritime decision-making scheme evaluation method integrating intuitionistic fuzzy as claimed in claim 1 is characterized in that: The specific operation of step S3 is as follows: Pseudo-subjective quantitative indicators, and calculate objective weights through entropy weight method: According to the evaluation matrix constructed in step S2, the quantitative indicators are pseudo-subjective calculated through intuitive fuzzy sets IFSs to obtain auxiliary quantitative indicators (pseudo-qualitative indicators), and the entropy weight method (Entropy Weight Method, EWM) is used to fuse the quantitative indicators and auxiliary quantitative indicators to achieve reasonable calculation of objective weights. The addition of auxiliary quantitative indicators makes the quantitative indicators subjective, so as to more effectively play the information value of quantitative indicators; Specifically, the dataset X = U is defined as the set of possible elements of the indicators considered in all maritime decision-making schemes. This dataset provides a complete and non-missing set of elements for the intuitionistic fuzzy set; Define an intuitionistic fuzzy set A. Based on the data set X, A can be expressed as: A = {(x, μ A (x),γ A (x))|x∈X}; Among them, μ A (x) is the membership function, which indicates the degree to which element x belongs to set A; γ A (x) is the non-membership function, which indicates the degree to which element x does not belong to set A; Among them, the membership degree μ A (U) and non-membership γ A (U) needs to be determined through expert knowledge or experience, or can be determined based on pre-set rules of expert knowledge; Through π A (x) = 1-μ A (x)-γ A (x) Calculate the uncertainty measure. The higher the value obtained, the higher the uncertainty about whether element x belongs to set A. The uncertainty measure quantifies the ambiguity of the affiliation of element x and expresses the uncertainty of the indicator in numerical form. A high uncertainty measure means that there is a greater hesitation or uncertainty in the classification of element x, that is, the pseudo-subjectivity of element x (i.e., quantitative indicator) is higher. Pseudo-subjectivity means that in the evaluation process, a certain degree of subjectivity is given to those quantitative indicators that are essentially objective through specific methods in order to better reflect the experience, intuition and preference of decision makers. The addition of pseudo-qualitative indicators makes quantitative indicators subjective, so as to more effectively exert the information value of quantitative indicators and achieve the purpose of fully mining the information of quantitative indicators. Define a mapping from a quantitative indicator to an intuitionistic fuzzy set A. Based on the membership and non-membership, convert the quantitative indicator into an element in the intuitionistic fuzzy set A. Calculate the objective weight of the quantitative indicator through the entropy weight method. Specifically, the entropy weight method is used to calculate the objective weight: Since the dimensions and numerical ranges of different indicators may be different, the original data needs to be standardized first; use range standardization Standardize the quantitative indicators and auxiliary quantitative indicators; where x ij is the original data of the jth indicator under the i-th scheme, x i ′ j is the standardized data, max(x i ) and min(x i ) are the maximum and minimum values ​​of all schemes of the ith indicator respectively; the range standardization method can simply, intuitively and quickly convert data in different dimensions and numerical ranges into a unified [0,1] interval, which is convenient for subsequent comparison and processing; Calculate the information entropy to measure the discrete degree of quantitative indicators and auxiliary quantitative indicators Among them, e j is the information entropy of the jth indicator, b is the number of solutions, p ij is the standardized data x i ′ j The normalized value of is calculated as The degree of dispersion is used to measure the diversity and uncertainty of the distribution between indicators. A low degree of dispersion means that the indicator is more dispersed within the possible range of values, while a high degree of dispersion means that the data values ​​tend to be concentrated. Taking the launch time and launch position indicators as an example, the time and position data with concentrated discrete degree areas are identified as repeated objective indicators and assigned higher discrete values; Calculate the coefficient of difference g used to measure the ability of the indicator to distinguish j =1-e j ; Among them, g j is the coefficient of variation of the jth indicator; the coefficient of variation is used to quantify the relative dispersion of indicator data, reflecting the comparison between the centralized trend and the discrete trend of the data set; Taking the launch time and launch position indicators as an example, if the launch time indicator has a higher coefficient of variation, it means that this indicator has a greater variation than the launch position indicator and has a stronger ability to distinguish different situations; Finally, the weight of each indicator is calculated according to the coefficient of variation The larger the weight, the stronger the distinguishing ability of the indicator is, and the more important it is to the objective matrix; j is the weight of the jth indicator, and c is the total number of indicators.

5. The subjective and objective balance maritime decision-making scheme evaluation method integrating intuitionistic fuzzy as claimed in claim 1 is characterized in that: The specific operation of step S4 is as follows: Step S4, using the Bayesian network to calculate the subjective weight of the qualitative indicators: First, a Bayesian network graph structure, i.e., a directed acyclic graph (DAG), is constructed; each node represents the score of a certain indicator in different schemes, and the edge represents the correlation between qualitative indicators; the network structure G can be expressed as: G = (V, E); where V is the node set and E is the edge set; Specifically, in order to analyze the inherent fuzziness and uncertainty in qualitative indicators, a conditional probability table (CPT) is defined for each variable in the network to describe the probability distribution of the corresponding variable when the value of the relationship variable is given; for variable X and its relationship variable P f (X), CPT can be expressed as: P(X|P f (X)); P(X|P f (X)) is the relation variable P f (X) the probability distribution of variable X under given conditions; The parameters in the CPT are learned using evaluation expert knowledge and relationship indicators; it should be noted that if the variable X has n possible values, its CPT will contain n probability values, which need to satisfy the normalization condition of probability: ∑ x P(X=x|P f (X)) = 1; where x traverses all possible values ​​of X; Specifically, the expectation-maximization (EM) algorithm is used to learn the network parameters; First, initialize a θ parameter; Next, with the current parameter θ t and observed variable X to infer latent variable x to obtain distribution P(x|X,θ t ), and calculate the log-likelihood estimate LogL(θ t |X,x)Expectation about x: Find the parameter that maximizes the expected likelihood, that is: θ t+1 = arg maxQ(θ|θ t ). Finally, repeat the above two steps until the set convergence threshold or number of iterations is reached to obtain the local or global optimal parameters; use the constructed Bayesian network for reasoning; use the learned conditional probability table P(X|P f (X)), through the formula The subjective weight is calculated; where X is the relationship variable of Y and P(E) is the probability of a given condition E.

6. The subjective and objective balance maritime decision-making scheme evaluation method integrating intuitionistic fuzzy as claimed in claim 1 is characterized in that: The specific operations of step S5 are as follows: The dynamic combination method is used to calculate the subjective and objective dynamic combination weights: Using the formula Calculate the dynamic combination weight; where W j is the comprehensive weight of the jth indicator, W j ′ To calculate the subjective weight of the jth indicator through the Ye Si network, W j ″ is the objective weight of the jth indicator calculated by the intuitionistic fuzzy entropy weight method, and c is the number of evaluation indicators. Compared with the traditional simple combination method, the formula uses normalization to dynamically balance the subjective and objective weights. Relatively high comprehensive weights are assigned to indicators with a large proportion, and relatively low comprehensive weights are assigned to indicators with a small proportion. This helps to avoid the problem of scale inconsistency in weight calculation, making weight comparison and interpretation more intuitive and fair, thereby reasonably representing the impact of each indicator on the overall decision in subjective and objective multi-criteria decision analysis problems such as the evaluation of maritime decision-making plans.

7. The subjective and objective balance maritime decision-making scheme evaluation method integrating intuitionistic fuzzy as claimed in claim 1 is characterized in that: The specific operation of step S6 is as follows: Step S6, obtaining a final score according to the combined weight, ranking the multiple sea-going decision-making schemes by their merits, and obtaining the evaluation results of the multiple sea-going decision-making schemes: Using weighted composite operation Get the comprehensive score of the sea decision-making plan; among them, R i is the comprehensive score of the i-th maritime decision-making plan, W j is the comprehensive weight of the jth indicator, x j is the score of the jth indicator, and c is the number of indicators; Execute the sorting algorithm according to the comprehensive score R i , sort all the sea-related decision-making options, and generate a list R={…,s i ,...}.

Images